LaserSoft Imaging

Da? auch ein kalibrierter Scanner farbstichige Bilder von farbstichigen Vorlagen liefert, ist klar. Es hat mich jedoch ?berrascht, da? beim Scannen von weitgehend farbstichfreien Dias das Scanresultat stark blau- / cyanstichig war, und ich viel M?he aufwenden mu?te, um mit Silverfast das Ergebnis an das Orginal anzugleichen.

Ich verwende Silverfast 5.5 f?r Nikon Coolscan IV LS40 ED als Plugin f?r Photoshop 6.0.1. Den Scanner habe ich zuvor mit dem mitgeliefertem Target kalibriert.
Als Arbeitsfarbraum habe ich Adobe RGB 1998 in Photoshop und Silverfast eingestellt, in Silverfast habe ich bei Scanner->intern Kalibrierung und ColorSync (mit gleichem Resultat) verwendet. (Intern->Monitor steht auf Automatisch.) Der Monitor ist zwar nicht 100% kalibriert, aber doch gut genug, um ihn als Ursache auszuschlie?en. Au?erdem ist der Farbstich allein durch Betrachtung der RGB-Werte ersichtlich.
Wenn ich die Kalibrierungsvorlage scanne, so erscheint diese sehr gut. (Wenn ich sie bei deaktivierter Kalibrierung oder mit Nikon Scan 3.1 scanne immerhin noch gut.)

Zun?chst hatte ich den Verdacht, da? der Scanner vielleicht abh?ngig vom Motiv einen Wei?abgleich durchf?hren w?rde, was nat?rlich die Kalibrierung wertlos machen w?rde. Dies scheint jedoch nicht der Fall zu sein, weil der Farbstich f?r s?mtliche Aufnahmen des Films trotz unterschiedlicher Farbverteilung auf den Bildern sehr ?hnlich ist. Bei meinen Tests fiel mir schlie?lich auf, da? Bilder anderer Filmtypen so wie das Kalibrierungstarget korrekt dargestellt wurden.
Der Filmtyp, der den Blaustich liefert, ist: Fujichrom Sensia 400 RH 135-36 DX, Emulsionsnummer 255408. Dagegen scheinen z.B. Filme vom Typ Fujichrom Sensia 100 RA korrekt gescannt zu werden.
Als n?chstes vermutete ich, die Ursache k?nnte in der spektralen Verteilung der Dichte des Films liegen. Daher besorgte ich mir aus dem Internet das Datenblatt des Films. Die dort gezeigten Dichtekurven sind jedoch mit denen anderer Filme vergleichbar.
Schlie?lich befiel mich der Verdacht, die Scannerfirmware oder die Software k?nnten auf Grund irgendwelcher (mir unklaren) Kriterien ?hnlich wie bei einem Negativfilm eine Orangemaskierung abziehen. (Die Invertierung unterbleibt allerdings.)
Wahrscheinlich ist es die Scannerfirmware, weil der Blaustich unabh?ngig von der verwendeten Software (Silverfast oder Nikon Scan) auftritt.
Die Dichteverteilung des Films kann ich mittlerweile mit hoher Wahrscheinlichkeit ausschlie?en, und zwar aus folgendem Grund:
Die letzte Aufnahme auf einem Film wurde vom Labor teilweise abgeschnitten. Wenn ich sie mit dem Filmhalter FH3 scanne, bleibt ein wei?er Streifen im Scan. Das Bild ist praktisch farbstichfrei. Ich habe versucht andere Aufnahmen des Films unter ?hnlichen Bedingungen zu scannen, jedoch ohne Erfolg.
Nat?rlich kann man auf andere Filmtypen ausweichen, solange jedoch die Ursache nicht gekl?rt ist, bleibt das Risiko, dass mit anderen Filmen ?hnliche Effekte auftreten.
Als Anlage zwei Bilder mit geringer Aufl?sung (die unerheblich zur Beurteilung des Farbstichs ist). Die Gesichter der Personen habe ich in Photoshop mit Weichzeichner unkenntlich gemacht. Bild ?ist.tif? ist ein Orginalscan mit Defaulteinstellungen, Bild ?soll.tif? die gleiche Aufnahme, jedoch habe ich in Silverfast die Farben so korrigiert, dass sie weitgehend dem Dia gleichen.

(Same text in english)
Blue-cyan cast scanning slides with calibrated scanner, depending on typ of film

That even a calibrated scanner will give you images with color cast, if the scanned image has a color cast, is natural. I was however surprised, that the result of scanning slides nearly without color cast contained intensive blue- cyan color cast, and that I had to work hard, to adjust the result with Silverfast, so it looked like the original.

I`m using Silverfast 5.5 with Nikon Coolscan IV LS40 ED as plugin for Photoshop 6.0.1. I`ve calibrated the scanner with the target, which ships with Silverfast.
As working color space I `m using Adobe RGB 1998 both in Photoshop and in Silverfast, in Silverfast I`ve tried both calibration and colorsync for scanner->intern (with same result). (The setting for intern->monitor is automatic.) Even so the monitor is not 100% calibrated, its good enough calibrated to exclude that that it might cause this result. Moreover you can see the color cast only by looking at the RGB values.
If I scan the Calibration target, it looks very good. (If I scan it with deactivated calibration or with Nikon Scan 3.1, its looks at last good.)
First I suspected, that the scanner might do a white adjustment depending on the contents of the slide, which would make calibration useless. But this seems not to be the case, because the color cast is quite similar for all pictures on the film, even so their color distribution is different. During my tests I noticed, that images of other types of films showed correctly like the calibration target.
The type of film causing the blue cast is: Fujichrom Sensia 400 RH 135-36 DX, emulsion number 255408. E.g. films of type Fujichrom Sensia 100 RA seem to be scanned correctly.
Next thing I suggested was, the reason could be the spectral distribution of the density of the film. So I downloaded the data sheet of the film. The curves of density shown there seem to be comparable with that of other films however.
At last I suspected, that the firmware of the scanner or the software could (for reasons which I don?t know) subtract an orange mask like for a negativ. (The image is not inverted however.)
Distribution of density of the film seem to be unlikely now for the following:
The last picture of a film had been cut of partly by the laboratory. If I scan it using FH3, there remains a white stripe in the scan. This picture is nearly without color cast. I `ve tried to scan other images of the film with similar conditions, but without success.
Of course you can use other types of films, but as long as the reason has not been found, there remains the risk, that similar effects could happen with other films.
As attachment two pictures with low resolution (which is insignificant for recognizing the color cast). The faces of the persons I `ve mad e unrecognizable in Photoshop with the blur tool. Picture ?ist.tif? is an original scan with default settings, picture ?soll.tif? the same slide, however I `ve adjusted the colors with Silverfast so, that they look quite similar to the slide.

(Sorry, kann hier im Formular keine tiff-Files mitschicken. Liefere diese jedoch auf Wunsch nach.)

i have some extra magneta cast when working with slightly exposed negative film.
still i think silverfast is best software and i think they are working on such problems now
check the installation page>>>http://www.silverfast.com/support/nikon-m-install-en.htm

Dear Anonymous,
that's true, and that "correctly re-install" info text has been updated recently.

Thanks for your tips, but its surely not an installation problem. Most images are scanned correctly and the behavior is not random. Its absolutely reproducible that certain images are always scanned wrong. Nor are these images over or under exposed. The images are not famous, I have made a lot betters, but that does not give the scanner the right not do work properly.
Please note: I DO NOT BLAME SILVERFAST FOR THAT!!
The result is the same with Nikon Scan or v**s**n. (With the exception, that only Silverfast allows calibration.)
What I find disappointing, is, that there is no response from Nikon.
Hey developers, support personal and managers from Nikon, is there anybody from you reading the Silverfast newsgroup?
(Ok, perhaps this topic should be under the group Nikon scanners, but that did not yet exist when I first posted the problem.)
Mr. Zahorsky from Silverfast asked me with private mail for more information. Had the files I sent you been helpful?
While I first believed, that the effect would depend on the type of film I used, I don?t believe that any longer, because I scanned meanwhile about a dozen images from that type of film correctly. That?s bad news, since it means, that the same problem could arise with every image of every type of film. It depends only on the contents of the image. While that?s more logical, its curious that I have a complete sequence of such images.
Since Nikon gives the interface documentation only to companies and not to individual developers, I cannot tell exactly how it works. (If Nikon could make an exception to me, please let me know.) Of course I could try to disassembly the firmware or I could write a filter driver for the USB port to monitor whats going on, but I am much too busy doing more important and more useful things for that. So I can build only my own theories on a few things I know. I know:
When you scan negatives, then the orange mask has to be removed by calculations and than the image has to be inverted. Both could be done by the scanner or by the scanning software. Because with every scanning software the user has to decide, if she wants to scan positives or negatives, probably everybody would belief, that this decision would trigger both. But if you read the Silverfast image forum carefully, you will know, that Silverfast had problems with negafix for the Nikon scanners, because Nikon insisted in removing the orange mask by the scanner firmware. From that it follows, that the removal of the orange mask was not triggered by the user interface. (Otherwise Silverfast had only to switch on positive mode.) That means, there is an automatic decision by the scanner. What I belief now, is that this automatism makes mistakes. But that?s naturally. How could an automatism distinguish a positive, which looks much like a negative, from a negative? So there must be mistakes. The only question is, what ?look like a negative? means. While I personally not agree that my images look like negatives, the scanner has an other opinion. Some of the wrong scanned images, I have taken in the archeological museum of Heraclion. Some contain ceramics. A lot of yellow and brown. You would even belief that the colors are scanned correctly as long as I don?t show you the slide and as long as you havened seen the original. But on the rest of the film I photographed some people. You will see with the first eye, that the colors are wrong. Even Nikons ICE3 tries to repair the colors.
What I don?t understand yet, is why other people don?t seem to have that problem. Is there an (by Nikon) adjustable threshold of the algorithm, which is misadjusted on my scanner?
While there might be good reasons to remove the orange mask in the scanner (that will make simple scanning software independent of the scanner), there is no reason not to trigger that from the UI. But, dear Nikon developers, if you belief that it must be done automatically, please note, that I will never mount negatives into slides. So you can always switch of the automatism, if the slide adapter is used.
It seems, that Nikon has build in a command now, which allows Silverfast to switch off the automatism for negafix. If that?s right, than Silverfast could (at least optionally) switch the automatism off also in positive mode.